inSights the EarthScope newsletter

EarthScope National Office College of Oceanic and Atmos. Sciences Oregon State University 104 COAS Admin. Bldg. Corvallis, OR 97331-5503 NEWS Inside this issue...

Integrating GPS & InSAR Along the San Andreas System

Edge-Driven Convection Beneath the Rio Grande Rift

EarthScope News

2010 IRIS Workshop in Snowbird, UT

EarthScope facilities are funded by the National Science Foundation and are being operated and maintained by UNAVCO Inc. and the Incorporated Research Institutions for Seismology with contributions from the U.S. Geological Survey and several national and international organizations. The newsletter is published by the EarthScope National Office at Oregon State University. The content of the newsletter material represents the views of the author(s) and not necessarily of the National Science Foundation.

inSights is a quarterly publication showcasing exciting scientific findings, developments, and news relevant to the EarthScope program. Contact earthscope@coas.oregonstate.edu to be added or deleted from the hardcopy mailing list; electronic copies are available at www. earthscope.org/publications/onsite. Editor: Jochen Braunmiller, OSU/EarthScope National Office, [email protected]. Production Editor: Charlotte Goddard, OSU/EarthScope National Office, [email protected]

IRIS Celebrates its 25th Anniversary at the 2010 Workshop

More than 230 scientists, including over 50 students and postdocs, gathered large datasets and arrays. The poster sessions included EarthScope June 9-11 at the Snowbird Resort in Utah for the 2010 Workshop and to research from the Transportable Array, Flexible Array experiments, celebrate the 25th anniversary of IRIS, a success story of extraordinary and the magnetotelluric (MT) array. Special Interest Groups (SIGs) and collaboration within the seismological community. Plenary sessions covered facility breakouts provided updates on current activities and discussions a wide range of topics. Prompted by the recent earthquakes in Haiti on exciting future directions. Several breakout sessions focused on and Chile, "The Science and Policy of Deadly Earthquakes" session Education and Outreach. During the dinner on June 9, workshop highlighted societal aspects of reducing loss and lessons learned from participants celebrated the first 25 years of IRIS while looking forward catastrophic disasters. Scientific sessions showed results from current to building on its foundation of facilitation, collaboration, and education. research on mantle dynamics and tremor and transient slip events. (www.iris.edu/hq/iris_workshop2010). ■ These fields have benefited from EarthScope USArray instrumentation, which has provided opportunities to develop new data-driven analysis techniques. Stimulating presentations from the realm of exploration seismology introduced new concepts on how to best exploit Danielle Sumy (LDEO) shows seismicity in the Sea of Cortez from the amphibious SCOOBA experiment. Photos: Rick Callender. inSights the EarthScope newsletter summer 2010

EarthScope NEWSNews Integrating GPS & InSAR to Resolve Stressing Rates Along the San Andreas System The absence of a major earthquake over the past three centuries along the southern San Andreas fault, a region home to over 10 million people and the site of large earthquakes in the past, provides ample motivation for studying the loading conditions of an active plate boundary. A comparison of strain rate maps of the western United States, produced by 15 research groups using primarily the same GPS measurements, reveals that modeled rates can differ by a factor of 5-8, with largest differences along the most active faults. EarthScope San Andreas interpretive workshop participants (www.earthscope.org/eno/parks) show New estimates of seismic hazard will rely on high Simple dislocation models predict that strain rate how the PBO-GPS station at CSU San Bernardino is resolution measurements of crustal deformation is proportional to slip rate divided by the locking moving relative to North America. Photo: Shelley Olds. and a secure understanding of strain rate derived depth, so even when using similar slip rates, ■ Register & apply for travel support from such measurements. The growing archive strain rates can differ by a factor of 2-3 because of by July 15 for the EarthScope of EarthScope’s Plate Boundary Observatory different fault depth assumptions. Institute on the Spectrum of Fault (PBO) GPS data is uniquely positioned to provide Slip Behaviors, October 11-14 in a large-scale perspective on plate boundary strain; A collaborative effort involving 15 research groups Portland, OR (www.earthscope. however, it cannot accurately resolve the highest is currently comparing large-scale plate boundary org/workshops/fault_slip10). strain rates near the most active faults. Integrating strain rate maps to establish best practices for GPS and InSAR velocities may be key to improving strain rate estimation (see online version for ■ Attend the EarthScope workshop strain rate accuracy and resolution - information participants and results). Two main conclusions can for interpretive professionals in that is critical for assessing seismic hazards. be drawn from comparing the community maps: the Yellowstone-Snake River Plain- 1) Nearly identical GPS datasets result in very Teton region, September 9-12 in Strain rate, which is typically greatest within 10-50 different maps, and 2) It is difficult to determine Jackson, WY. Information and an km of an active fault, is calculated by taking the which maps capture the true strain rate best. application form (deadline August spatial derivative of measured crustal velocities Maps derived from isotropic interpolation suggest 1) are at www.earthscope.org/ (Figure 1). Full resolution of velocity gradients lower rates (e.g., 50-500 nanostrain/year, Imperial workshops/yellowstone. requires a spatial sampling at about 1/4 of the fault) than maps generated from dislocation typical 6-18 km fault locking depth, which is less models and methods that localize strain onto faults ■ Proposals for the next EarthScope than the typical ~10 km spacing of the present- (e.g., 1000-2700 nanostrain/year, Imperial fault). National Office are due October 1. day PBO network along the San Andreas system. The large variations are mainly due to different A physical model or an interpolation method methods used to construct a high resolution ■ The EarthScope Speaker Series must, therefore, be used to compute continuous presents EarthScope science at velocities before obtaining a strain rate map. continued on page 3 college and university seminars. To apply for a speaker visit www. 20 earthscope.org/speakers. San 10 Francisco 0 −10 velocity ■ EarthScope was selected as one −20 (mm/yr) 2000 of 15 NSF exhibits for the inaugural 1500 USA Science and Engineering 1000 500 strain rate (nanostrain/yr) Festival on Washington, DC’s San Andreas Fault 0 -100 km -50 km 0 km 50 km 100 km National Mall October 23-24 (www. usasciencefestival.org). More than 500 organizations will present Los Angeles hands-on activities to inspire the next generation of scientists.

■ The EarthScope Automated Receiver Survey (EARS) is now 2nd invariant strain rate (nanostrain/yr) San Diego implemented at the IRIS DMC 1 10 100 1000 (http://ears.iris.washington.edu). Developed by the University of Figure 1: Strain rate of the San Andreas Fault System from a geodetically constrained analytical model. Deep slip occurs on 41 fault segments where geologic slip rate is applied and locking depth is varied along each fault segment to best fit the GPS data. Dashed white line shows location of inset. continued on page 3 Inset: cross-section across the Imperial fault. Top: velocity profile (black line) and GPS data (gray circles). Bottom: strain rate across the Imperial fault. 2 inSights the EarthScope newsletter

Edge-Driven Convection Beneath the Rio Grande Rift The Rio Grande Rift is a series of north-south trending faulted basins extending for more than 1000 km from Colorado to Chihuahua, Mexico and the Big Bend region of Texas. The rift is a Cenozoic feature with a mid-Oligocene (~30 Ma) early rifting stage, possibly related to foundering of the flat-subducting Farallon plate, and a recent late Miocene (~10 Ma) phase, which continues today.

There is no clear cause for the resurgence thickened lithosphere beneath the Great Plains flow induced by the edge convection or even in magmatism and extension. However, the and fast velocities in the mantle’s top 500 km delamination of lithosphere. Geodynamic rift location at the edge of the Great Plains, that might indicate thermal and/or density modeling constrained by SIEDCAR will help with its stable, thick lithosphere suggests anomalies. If these anomalies are detected, a determine whether this is happening at the that edge-driven convection along the eastern second goal is to obtain quantitative estimates boundary between the Great Plains and the border of the Southern Rockies may play an of their size, geometry, and contrasts for use Rio Grande Rift. important role in the current tectonics of the in geodynamic modeling. Rio Grande Rift. In particular, edge-driven Edge-driven convection may explain why, after convection may erode the lithosphere beneath Initial results of SKS splitting measurements a long period of quiescence, the Rio Grande the Rift and Great Plains, cause lower crustal to constrain patterns of mantle anisotropy Rift became active again. The SIEDCAR flow, and produce magmatism long after (Figure 1), receiver functions to show crustal project seeks to understand how this process initiation of the rift. thickness (Figure 2), and body-wave and continues to modify the sub-continental surface-wave tomography to determine P and lithosphere long after the formation of the In 2008, a group of university scientists and S wave velocities show patterns consistent rift. ■ high school teachers deployed 71 broadband with edge-driven convection. Tomographic seismographs interspersed between the images indicate that the fast upper mantle By Jay Pulliam a, Stephen P. Grand b, Yu Xia b, Carrie Rockett a, EarthScope Transportable Array (TA) stations, anomaly beneath the eastern flank of the and Tia Barrington a, a Baylor University, b University of Texas essentially doubling the station density in Rio Grande Rift, which was identified by the at Austin. southeastern New Mexico and west Texas earlier, linear, La Ristra array, (Figure 1). is clearly separated from Rio Grande Rift Great Plains 107° 106° 105° 104° 103° 102° the Great Plains craton and thinner crust thicker crust Albuquerque Amarillo that it extends southward 0 35° to at least the Big Bend 20 region of Texas. SKS 40 Socorro splitting measurements 60

34° show a marked decrease depth (km) 80 Rio Lubbock in splitting times above the Grande 100 Rift fast “downwelling” anomaly -106 -105.5 -105 -104.5 -104 -103.5 longitude 33° in the mantle and generally Great Figure 2: Receiver function image of crust and uppermost mantle beneath the eastern Plains larger splits on the rift flank Las Cruces flank of the Rio Grande Rift from SIEDCAR and TA data for the NW-SE oriented La proper. Receiver function Midland Ristra transect (dotted line in Figure 1). Thickened crust beneath the Great Plains 32° images suggest a thickening extends both to the north and to the south, but, south of this cross-section, the El Paso Odessa of the crust to the east, which transition from thick to thin crust moves to the southwest. might indicate lower crustal 31°

−124˚ −120˚ −116˚ −112˚ −108˚ −104˚ 30° EarthScope CASCADIA ETS Active Flex Array Stations Array of Arrays 48˚ (Past Flex Array Stations) 0.5s 1.0s1.5s 2.0s Temporary Campaign GPS Stations (Cafe) SNAKE RIVER (Wallowa) Figure 1: SKS splitting parameters for SIEDCAR (red) and TA Deployments PLAIN (green) stations deployed 2008-2010 in the study area. Previous FACES splitting measurements (La Ristra linear array and others) are COLZA Temporary deployments of seismic and geodetic shown in blue. Dash-dot circle indicates area where the SKS 44˚ instruments allow for focused observation and delay-times decrease; this area coincides with the “downwelling” Bighorn study of key geophysical targets. EarthScope fast anomaly in the mantle. The NW-SE-oriented La Ristra line is maintains ~100 Campaign GPS systems and indicated by a dotted line. ~2100 broadband and short-period FlexArray (NW Nevada) The dense network increases resolution of seismic sensors available for PI-driven, focused 40˚ studies. This map shows present and past tomographic and receiver function images (SNEP) RIO seismic and GPS deployments. The article (Mendocino) GRANDE and shear wave splitting measurements, on this page highlights initial results from and allows construction of a detailed 3-D SIEDCAR, one of the current studies. The investigation of the eastern edge of the Rio Sierra Nevada EarthScope Project (SNEP) was 36˚ COLORADO PLATEAU Grande Rift structure. The primary goal of featured in the spring/summer 2008 onSite, SIEDCAR SAN BERNARDINO SIEDCAR (Seismic Investigation of Edge while the Rio Grande and the Snake River Plain SAF Driven Convection Associated with the Rio GPS experiments were highlighted in the winter 2009 onSite. For more information on field Grande Rift) is to determine if the necessary 32˚ km experiments visit www.earthscope.org/science/ 0 125 250 conditions for small-scale convection exist field_programs. over a broad area. The conditions include inSights the EarthScope newsletter 3

after the April 4, 2010 M 7.2 northern Baja students, teachers, faculty, news media, California earthquake. Four EarthScope and the public, can be accessed at www. EarthScope News campaign GPS receivers were used earthscope.org/eno. to provide geodetic control for the TLS continued from front ■ The Bighorns Flexible Array NEWS surveys. For more information visit www. experiment recently deployed more South Carolina as an EarthScope data product, unavco.org/research_science/science_ than 150 short period instruments in EARS calculates bulk crustal properties at USArray highlights/2010/M7.2-Baja.html. the Bighorn Mountains of Wyoming. sites using receiver functions. Visit www.iris.edu/ ■ The revised and expanded EarthScope The experiment uses newly developed dms/products/ears for more information on EARS Education and Outreach web pages "quick deploy" boxes where stations are and www.earthscope.org/science/data_products include links to handouts, animations, shipped in deployment-ready enclosures. for other EarthScope data products. teachable moments, presentations They worked so well that the field teams ■ UNAVCO provided a terrestrial laser scanner (TLS, and much more from EarthScope and completed station installations one ground based LiDAR) to scan surface ruptures its partners. The materials, aimed at week ahead of schedule.

continued from front Integrating GPS & InSAR to Resolve Stressing Rates Along the SAF System (1 km) map from sparse (~10 km) GPS data. Second order differences can coherence is poor at C-band, will require using the longer be attributed to assumed fault locations and rheological assumptions. wavelength L-band data from ALOS.

How can the geodetic community improve accuracy and resolution The likelihood of a major earthquake depends on the of strain rate maps? First, when GPS resolution is inadequate, proper accumulated stress, which is roughly equal to strain rate localization of strain seems to require that models include major fault multiplied by the crustal shear modulus and the time since the locations. Second, a solution is to densify the GPS network such that last major rupture. Strain rate mapping is, therefore, only one station spacing is smaller than the spatial variations in crustal strain. component needed to forecast earthquakes; paleoseismic This approach has been tested in areas such as the Imperial fault, where Model + Residuals = Final Interferogram strain is adequately resolved. Indeed, installation of a few dense GPS mm/yr 34˚ −3 0 3 arrays across poorly-resolved, high-strain-rate faults is feasible and should SAF be pursued. Recently we deployed a second dense GPS array across the Imperial fault to resolve the high strain rate in the highly-populated SAF SAF Mexicali area of Baja California p2 Salton Sea p2 (Figure 2).

33˚ SHF SHF p1 SHF p1 Imperial Fault (37 mm/yr) Figure 2: Interferogram of the April 2010 a b mm/yr c Mexicali −116˚ −115˚ −3 0 3 earthquake in Baja California, Mexico, from ALOS Baja California PALSAR (Advanced Land Observing Satellite 5 SHF 10 SAF NE -10 0 10 Laguna mainSalada rupture LOS velocity (mm/yr) Phased Array type L-band Synthetic Aperture p1 p2 Radar). One color cycle is 11.6 cm of line of sight (4 mm/yr) 0 0 Model mm/yr InSAR deformation. Locations of dense GPS monuments Cerro Prieto (37 mm/yr) SW −5 −10 across the Imperial fault are also shown. The liquefaction zone d −20 −10 0 10 −60 −40 −20 0 20 40 60 Imperial array (yellow) was first surveyed in 1993. km km Reoccupations in 1999, 2000 and 2008 have Figure 3: Remove-stack-restore interferograms for the Salton Sea area. (a) Physical model provided unprecedented accuracy and resolution. constrained by GPS data. (b) Stacked residual interferograms after applying the filtering In January 2010, 17 new monuments (red) were method. (c) Final interferogram; sum of (a) and (b). Gray solid lines in (a) and (c) correspond installed across the fault in Mexicali. to profile locations in (d). White arrow indicates satellite look direction. (d) Profile across Superstition Hills fault (SHF) and the main faults of the southern San Andreas fault (SAF) system. Arrow identifies short wavelength signal absent in GPS data. A third approach is to utilize InSAR’s spatial coverage and to stack long time interval interferograms to augment GPS derived estimates. We are estimates of the timing and slip of recent (~2000 years) major developing a remove-stack-restore procedure to optimally combine GPS ruptures are equally important. It is interesting to note that the and InSAR, which considers signal, measurement noise, sampling rate, last four major earthquakes in Southern California have not and environmental noise characteristics of each system (Figure 3). This occurred on faults with the highest slip rate (Landers, 1992; method involves constructing a wide-area, 3-D dislocation model from the Northridge, 1994; Hector Mine, 1999; and the recent Sierra El GPS data; after low-pass filtering at 40 km, the model is removed from Mayor-Cucapah earthquake [Figure 2]). In addition to resolving each interferogram. Because the residual interseismic signal and noise high strain rates on the primary faults, it is thus very important have different scale dependencies, filtering an interferogram can increase to measure the lower rates on subsidiary faults that rupture the signal-to-noise ratio by as much as 20%. Multiple interferograms are far less often but have recently produced the most damaging stacked to reduce atmospheric error. Finally, the complete deformation events. GPS and InSAR, when optimally integrated, will be the field is constructed by adding the low-pass filtered model back to the primary geodetic tools for resolving crustal strain rate in the stack of interferograms. Application to a large stack of ERS (European most critical regions of our active plate boundary. ■ Remote Sensing) interferograms in Southern California demonstrates better than 2 mm/yr accuracy over the wide range of length scales (200 By Bridget Smith-Konter a, David T. Sandwell b, and Meng Wei b, a University of Texas m to 500 km). Extending the method to Northern California, where InSAR at El Paso, b Scripps Institution of Oceanography.